[
https://issues.apache.org/jira/browse/FLINK-32974?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Lijie Wang updated FLINK-32974:
-------------------------------
Description:
After FLINK-32226, a temporary directory(named
{{flink-rest-client-jobgraphs*}}) is created when creating a new
RestClusterClient, but this directory will never be cleaned up.
This will result in a lot of {{flink-rest-client-jobgraphs*}} directories under
{{/tmp}}, especially when using CollectDynamicSink/CollectResultFetcher, which
may cause the inode to be used up.
was:
After FLINK-32226, a temporary directory(named
{{flink-rest-client-jobgraphs*}}) is created when creating a new
RestClusterClient, but this directory will never be cleaned up.
This will result in a lot of {{flink-rest-client-jobgraphs*}} directories under
{{/tmp}}, especially when using CollectDynamicSink/CollectResultFetcher.
> RestClusterClient always leaks flink-rest-client-jobgraphs* directories
> -----------------------------------------------------------------------
>
> Key: FLINK-32974
> URL: https://issues.apache.org/jira/browse/FLINK-32974
> Project: Flink
> Issue Type: Bug
> Components: Client / Job Submission
> Affects Versions: 1.18.0, 1.17.2
> Reporter: Lijie Wang
> Assignee: Lijie Wang
> Priority: Critical
>
> After FLINK-32226, a temporary directory(named
> {{flink-rest-client-jobgraphs*}}) is created when creating a new
> RestClusterClient, but this directory will never be cleaned up.
> This will result in a lot of {{flink-rest-client-jobgraphs*}} directories
> under {{/tmp}}, especially when using
> CollectDynamicSink/CollectResultFetcher, which may cause the inode to be used
> up.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
